## [1] "C:/Users/kadwolf/OneDrive - UGent/UGent-PC/Kadwolf/Documents/00_SPINCITY/05_onderzoek/03_colour_analysis/spin_city_colour_results_V1"
analysis data collected in 2021
SAMPLING INFO CONTAINS 85 sampling sites: - 2 repeated samplings because the spiders were not frozen immediately: P05SR_R and P04SG_R (i remove them in the colour analysis as it is then the most randomized scheme and otherwise it will increase sampling sizes of location P05SR_R and P04SG_R with 20 spiders each) - 1 extra location in Brussels P20SE (RBINS) => i indicated it for now as High Medium urbanisation category (check with urb radius) - 1 extra location in Ghent, P01SE, but we did not collect spiders
to conclude : all 81 typical speedy sampling sites were sampled; at P09SY no A. diadematus found
=> for this analysis i remove location P05SR_R P04SG_R P20SE:
## [1] "P01SG" "P01SR" "P01SY" "P02SG" "P02SR" "P02SY" "P03SG"
## [8] "P03SR" "P03SY" "P04SG_R" "P04SG" "P04SR" "P04SY" "P05SG"
## [15] "P05SR_R" "P05SR" "P05SY" "P06SG" "P06SR" "P06SY" "P07SG"
## [22] "P07SR" "P07SY" "P08SG" "P08SR" "P08SY" "P09SG" "P09SR"
## [29] "P10SG" "P10SR" "P10SY" "P11SG" "P11SR" "P11SY" "P12SG"
## [36] "P12SR" "P12SY" "P13SG" "P13SR" "P13SY" "P14SG" "P14SR"
## [43] "P14SY" "P15SG" "P15SR" "P15SY" "P16SG" "P16SR" "P16SY"
## [50] "P17SG" "P17SR" "P17SY" "P18SG" "P18SR" "P18SY" "P19SG"
## [57] "P19SR" "P19SY" "P20SE" "P20SG" "P20SR" "P20SY" "P21SG"
## [64] "P21SR" "P21SY" "P22SG" "P22SR" "P22SY" "P23SG" "P23SR"
## [71] "P23SY" "P24SG" "P24SR" "P24SY" "P25SG" "P25SR" "P25SY"
## [78] "P26SG" "P26SR" "P26SY" "P27SG" "P27SR" "P27SY"
preparation of dataset:
## [1] "spiderID" "a" "r"
## [4] "col_corr_leafdark" "col_corr_leaflight" "col_corr_cross"
## [7] "col_corr_abdbri" "spider_length" "abdomen_length"
## [10] "cross_length" "cross_width" "abdomen_area"
## [13] "project_year" "location" "plotid"
## [16] "city" "region" "U_landscape"
## [19] "U_local" "urb_cat" "sampling_date"
## [22] "project_detail" "remark_sampling" "x"
## [25] "y"
## spc_tbl_ [1,383 × 25] (S3: spec_tbl_df/tbl_df/tbl/data.frame)
## $ spiderID : chr [1:1383] "SC21P01SG01" "SC21P01SG02" "SC21P01SG03" "SC21P01SG04" ...
## $ a : num [1:1383] 8.96 8.61 9.03 9.3 9.82 ...
## $ r : num [1:1383] 0.0122 0.0108 0.0119 0.0121 0.0111 ...
## $ col_corr_leafdark : num [1:1383] 12.8 17.4 12.3 21.2 25.9 ...
## $ col_corr_leaflight: num [1:1383] 20.5 21.9 15 31.6 29 ...
## $ col_corr_cross : num [1:1383] 60.1 49.5 75.6 61.1 47.1 ...
## $ col_corr_abdbri : num [1:1383] 19.4 22.1 18.9 29.1 38.4 ...
## $ spider_length : num [1:1383] 13.3 16.3 14.6 13.3 18 ...
## $ abdomen_length : num [1:1383] 10.7 13.5 12.7 11.8 15.9 ...
## $ cross_length : num [1:1383] 5.38 6.63 5.22 5.11 7.45 ...
## $ cross_width : num [1:1383] 4.29 4.94 5.01 3.3 5.24 ...
## $ abdomen_area : num [1:1383] 80.7 128.7 104.3 91.2 164.7 ...
## $ project_year : chr [1:1383] "SC21" "SC21" "SC21" "SC21" ...
## $ location : chr [1:1383] "P01SG" "P01SG" "P01SG" "P01SG" ...
## $ plotid : chr [1:1383] "P01" "P01" "P01" "P01" ...
## $ city : chr [1:1383] "Gent" "Gent" "Gent" "Gent" ...
## $ region : chr [1:1383] "Gent" "Gent" "Gent" "Gent" ...
## $ U_landscape : chr [1:1383] "HIGH" "HIGH" "HIGH" "HIGH" ...
## $ U_local : chr [1:1383] "LOW" "LOW" "LOW" "LOW" ...
## $ urb_cat : chr [1:1383] "HIGHLOW" "HIGHLOW" "HIGHLOW" "HIGHLOW" ...
## $ sampling_date : chr [1:1383] "27/09/2021" "27/09/2021" "27/09/2021" "27/09/2021" ...
## $ project_detail : chr [1:1383] "staalname2021" "staalname2021" "staalname2021" "staalname2021" ...
## $ remark_sampling : chr [1:1383] "met bram" "met bram" "met bram" "met bram" ...
## $ x : chr [1:1383] "51.049036" "51.049036" "51.049036" "51.049036" ...
## $ y : chr [1:1383] "3.696775" "3.696775" "3.696775" "3.696775" ...
## - attr(*, "spec")=
## .. cols(
## .. spiderID = col_character(),
## .. a = col_double(),
## .. r = col_double(),
## .. col_corr_leafdark = col_double(),
## .. col_corr_leaflight = col_double(),
## .. col_corr_cross = col_double(),
## .. col_corr_abdbri = col_double()
## .. )
## - attr(*, "problems")=<externalptr>
## [1] "P01SG" "P01SR" "P01SY" "P02SG" "P02SR" "P02SY" "P03SG" "P03SR" "P03SY"
## [10] "P04SG" "P04SR" "P04SY" "P05SG" "P05SR" "P05SY" "P06SG" "P06SR" "P06SY"
## [19] "P07SG" "P07SR" "P07SY" "P08SG" "P08SR" "P08SY" "P09SG" "P09SR" "P10SG"
## [28] "P10SR" "P10SY" "P11SG" "P11SR" "P11SY" "P12SG" "P12SR" "P12SY" "P13SG"
## [37] "P13SR" "P13SY" "P14SG" "P14SR" "P14SY" "P15SG" "P15SR" "P15SY" "P16SG"
## [46] "P16SR" "P16SY" "P17SG" "P17SR" "P17SY" "P18SG" "P18SR" "P18SY" "P19SG"
## [55] "P19SR" "P19SY" "P20SG" "P20SR" "P20SY" "P21SG" "P21SR" "P21SY" "P22SG"
## [64] "P22SR" "P22SY" "P23SG" "P23SR" "P23SY" "P24SG" "P24SR" "P24SY" "P25SG"
## [73] "P25SR" "P25SY" "P26SG" "P26SR" "P26SY" "P27SG" "P27SR" "P27SY"
## [1] "HIGH" "LOW" "MEDIUM"
## [1] "HIGH" "LOW" "MEDIUM"
## [1] "HIGHHIGH" "HIGHLOW" "HIGHMEDIUM" "LOWHIGH" "LOWLOW"
## [6] "LOWMEDIUM" "MEDIUMHIGH" "MEDIUMLOW" "MEDIUMMEDIUM"
## Warning: package 'chron' was built under R version 4.3.2
## [1] TRUE
exploration of data:
## Rows: 1,383
## Columns: 26
## $ spiderID <fct> SC21P01SG01, SC21P01SG02, SC21P01SG03, SC21P01SG04,…
## $ a <dbl> 8.959275, 8.610263, 9.033979, 9.297104, 9.822551, 6…
## $ r <dbl> 0.01222899, 0.01077165, 0.01193368, 0.01212227, 0.0…
## $ col_corr_leafdark <dbl> 12.772954, 17.404049, 12.271638, 21.200340, 25.9076…
## $ col_corr_leaflight <dbl> 20.49506, 21.90025, 15.03173, 31.62829, 28.96291, 1…
## $ col_corr_cross <dbl> 60.11953, 49.47924, 75.57975, 61.11121, 47.12496, 4…
## $ col_corr_abdbri <dbl> 19.44479, 22.07536, 18.89618, 29.11122, 38.44574, 1…
## $ spider_length <dbl> 13.346, 16.261, 14.619, 13.343, 17.953, 16.827, 15.…
## $ abdomen_length <dbl> 10.676, 13.489, 12.739, 11.823, 15.852, 13.401, 12.…
## $ cross_length <dbl> 5.378, 6.629, 5.224, 5.113, 7.448, 5.771, 4.982, 5.…
## $ cross_width <dbl> 4.291, 4.939, 5.013, 3.297, 5.242, 3.774, 3.649, 5.…
## $ abdomen_area <dbl> 80.699, 128.657, 104.282, 91.243, 164.744, 118.947,…
## $ project_year <fct> SC21, SC21, SC21, SC21, SC21, SC21, SC21, SC21, SC2…
## $ location <fct> P01SG, P01SG, P01SG, P01SG, P01SG, P01SG, P01SG, P0…
## $ plotid <fct> P01, P01, P01, P01, P01, P01, P01, P01, P01, P01, P…
## $ city <fct> Gent, Gent, Gent, Gent, Gent, Gent, Gent, Gent, Gen…
## $ region <fct> Gent, Gent, Gent, Gent, Gent, Gent, Gent, Gent, Gen…
## $ U_landscape <fct> HIGH, HIGH, HIGH, HIGH, HIGH, HIGH, HIGH, HIGH, HIG…
## $ U_local <fct> LOW, LOW, LOW, LOW, LOW, LOW, LOW, LOW, LOW, LOW, L…
## $ urb_cat <fct> HIGHLOW, HIGHLOW, HIGHLOW, HIGHLOW, HIGHLOW, HIGHLO…
## $ sampling_date <date> 2021-09-27, 2021-09-27, 2021-09-27, 2021-09-27, 20…
## $ project_detail <chr> "staalname2021", "staalname2021", "staalname2021", …
## $ remark_sampling <chr> "met bram", "met bram", "met bram", "met bram", "me…
## $ x <chr> "51.049036", "51.049036", "51.049036", "51.049036",…
## $ y <chr> "3.696775", "3.696775", "3.696775", "3.696775", "3.…
## $ day <dbl> 269, 269, 269, 269, 269, 269, 269, 269, 269, 269, 2…
## [1] "cross_length" "cross_width" "remark_sampling"
| plotid | U_landscape | LOW | MEDIUM | HIGH |
|---|---|---|---|---|
| P01 | HIGH | 20 | 18 | 20 |
| P02 | HIGH | 22 | 17 | 20 |
| P03 | HIGH | 21 | 11 | 19 |
| P04 | MEDIUM | 20 | 20 | 17 |
| P05 | MEDIUM | 8 | 22 | 20 |
| P06 | MEDIUM | 20 | 20 | 16 |
| P07 | LOW | 14 | 19 | 24 |
| P08 | LOW | 19 | 17 | 19 |
| P09 | LOW | 19 | NA | 20 |
| P10 | HIGH | 20 | 12 | 17 |
| P11 | HIGH | 20 | 20 | 14 |
| P12 | HIGH | 20 | 20 | 21 |
| P13 | MEDIUM | 12 | 20 | 19 |
| P14 | MEDIUM | 12 | 9 | 20 |
| P15 | MEDIUM | 10 | 18 | 18 |
| P16 | LOW | 19 | 1 | 20 |
| P17 | LOW | 17 | 5 | 21 |
| P18 | LOW | 18 | 12 | 14 |
| P19 | HIGH | 20 | 17 | 20 |
| P20 | HIGH | 14 | 9 | 15 |
| P21 | HIGH | 19 | 19 | 19 |
| P22 | MEDIUM | 20 | 15 | 22 |
| P23 | MEDIUM | 20 | 21 | 21 |
| P24 | MEDIUM | 20 | 19 | 22 |
| P25 | LOW | 16 | 2 | 15 |
| P26 | LOW | 8 | 19 | 17 |
| P27 | LOW | 22 | 21 | 20 |
| U_landscape | LOW | MEDIUM | HIGH |
|---|---|---|---|
| LOW | 152 | 96 | 170 |
| MEDIUM | 142 | 164 | 175 |
| HIGH | 176 | 143 | 165 |
remove those individuals so we get dataset : data_bz to analyse the morphological measurements
first exploration then test each morphological trait via univariate mixed models
always the same 3 type of visualisations are repeated
maybe better to scale the cross length to the proportion it takes of the abdomen or of the spider visualised here
## Warning: package 'ggpubr' was built under R version 4.3.2
as total spider length was less accurate to measure, positioning spider plays a role, abdomen corrected is here most suitable i think
both show rougly the same trends
###statistics
statistics make use of glmmTMB multiple observation for each subplotid
first check correlation of sampling day with the body size meaurements. However only very minor correlation present
## Warning: package 'corrplot' was built under R version 4.3.2
that’s why i don’t include day (scaled sampling day) in the statistical models
## Warning: package 'DHARMa' was built under R version 4.3.3
## Warning: package 'glmmTMB' was built under R version 4.3.3
## Warning: package 'car' was built under R version 4.3.2
## Warning: package 'carData' was built under R version 4.3.2
## Warning: package 'emmeans' was built under R version 4.3.3
## Warning: package 'performance' was built under R version 4.3.3
## Warning: package 'effects' was built under R version 4.3.2
## Family: gaussian ( identity )
## Formula: spider_length ~ U_landscape + U_local + (1 | plotid/location)
## Data: data_bz
##
## AIC BIC logLik deviance df.resid
## 5565.1 5606.9 -2774.5 5549.1 1368
##
## Random effects:
##
## Conditional model:
## Groups Name Variance Std.Dev.
## location:plotid (Intercept) 0.3814 0.6176
## plotid (Intercept) 0.3430 0.5857
## Residual 3.0320 1.7413
## Number of obs: 1376, groups: location:plotid, 80; plotid, 27
##
## Dispersion estimate for gaussian family (sigma^2): 3.03
##
## Conditional model:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 13.165865 0.272824 48.26 <2e-16 ***
## U_landscapeMEDIUM -0.342033 0.347725 -0.98 0.325
## U_landscapeHIGH 0.545934 0.347465 1.57 0.116
## U_localMEDIUM 0.006546 0.212754 0.03 0.975
## U_localHIGH 0.234686 0.202906 1.16 0.247
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## Analysis of Deviance Table (Type III Wald chisquare tests)
##
## Response: spider_length
## Chisq Df Pr(>Chisq)
## (Intercept) 2328.8124 1 < 2e-16 ***
## U_landscape 6.8145 2 0.03313 *
## U_local 1.7082 2 0.42567
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## contrast estimate SE df t.ratio p.value
## LOW - MEDIUM 0.342 0.348 1368 0.984 0.5873
## LOW - HIGH -0.546 0.347 1368 -1.571 0.2585
## MEDIUM - HIGH -0.888 0.343 1368 -2.588 0.0264
##
## Results are averaged over the levels of: U_local
## P value adjustment: tukey method for comparing a family of 3 estimates
random factor: (1|plotid/location) : plotid (P01-P27) and location (P01SG, .., P27SR) fixed effects: urbanisation levels at landscape and local scale : U_landscape and U_local
Result: spider length : spiders become bigger with increasing urbanisation level at landscape scale, but only difference between medium and high urbanised landscapes are significant
spider length sometimes difficult to accurately measure. maybe abdomen length is somehow measured more comparable
## Family: gaussian ( identity )
## Formula:
## abdomen_length ~ U_landscape + U_local + (1 | plotid/location)
## Data: data_bz
##
## AIC BIC logLik deviance df.resid
## 5503.0 5544.8 -2743.5 5487.0 1368
##
## Random effects:
##
## Conditional model:
## Groups Name Variance Std.Dev.
## location:plotid (Intercept) 0.4397 0.6631
## plotid (Intercept) 0.3122 0.5588
## Residual 2.8823 1.6977
## Number of obs: 1376, groups: location:plotid, 80; plotid, 27
##
## Dispersion estimate for gaussian family (sigma^2): 2.88
##
## Conditional model:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 10.92572 0.27230 40.12 <2e-16 ***
## U_landscapeMEDIUM -0.47067 0.34337 -1.37 0.170
## U_landscapeHIGH 0.14240 0.34312 0.42 0.678
## U_localMEDIUM 0.09656 0.22149 0.44 0.663
## U_localHIGH 0.21502 0.21183 1.02 0.310
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## Analysis of Deviance Table (Type III Wald chisquare tests)
##
## Response: abdomen_length
## Chisq Df Pr(>Chisq)
## (Intercept) 1609.9318 1 <2e-16 ***
## U_landscape 3.5815 2 0.1668
## U_local 1.0348 2 0.5961
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
###abdomen area - statistics abdomen area as a potential measure of fecundity
## Family: gaussian ( identity )
## Formula: abdomen_area ~ U_landscape + U_local + (1 | plotid/location)
## Data: data_bz
##
## AIC BIC logLik deviance df.resid
## 12901.4 12943.2 -6442.7 12885.4 1368
##
## Random effects:
##
## Conditional model:
## Groups Name Variance Std.Dev.
## location:plotid (Intercept) 97.46 9.872
## plotid (Intercept) 77.09 8.780
## Residual 621.92 24.938
## Number of obs: 1376, groups: location:plotid, 80; plotid, 27
##
## Dispersion estimate for gaussian family (sigma^2): 622
##
## Conditional model:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 80.470 4.152 19.380 <2e-16 ***
## U_landscapeMEDIUM -7.313 5.273 -1.387 0.165
## U_landscapeHIGH 2.330 5.269 0.442 0.658
## U_localMEDIUM 1.097 3.285 0.334 0.738
## U_localHIGH 2.553 3.142 0.812 0.417
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## Analysis of Deviance Table (Type III Wald chisquare tests)
##
## Response: abdomen_area
## Chisq Df Pr(>Chisq)
## (Intercept) 375.5914 1 <2e-16 ***
## U_landscape 3.7287 2 0.1550
## U_local 0.6651 2 0.7171
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
no differences in abdomen area
check for strange values although the colour correction is executed. In quite a lot of pictures the grey scale is blurry and in some also the spider
###exploration
## Min. 1st Qu. Median Mean 3rd Qu. Max.
## 4.905 18.885 21.826 22.727 25.900 77.857
SC21P10SR15 => spider is pale but not that pale also not that pale :
SC21P24SR05, SC21P22SR18 SC21P21SG12 => is not that dark spider
dataset for the colour analysis : data_col data_col contain 1380 spiders! (instead of 1383 spider, data_bz contains 1376 spiders)
###correlation
no trends with day (sampling day). However, there is a strong positive
correlation between abdomen brightness and leafdark, abdomen brightness
and leaflight => correlation: the paler the spider also the
paler the leafdark and leaflight
correlation: the paler the leafdark also the paler the
leaflight
no correlation with colour of the cross these values are still strange although correction was applied in the same way as for col_corr_leafdark and col_corr_leafdark => maybe only focus on % reflectance of abdomen !
Again 3 type of visualisation graphs and after that the statistical test
the same dataset but now with some more strict filtering of extremes
(probably due to photo quality) so everything under 10 and everything
above 35 => dataset : data_col_filt: contains 1337
observations(=spiders)
## [1] 1379 27
## [1] 1337 27
the same 3 graph visualisations for the filtered dataset + violin
###statistics ###statistics - data_col statistics on abdomen brightness: now expressed as reflectance
## Family: gaussian ( identity )
## Formula:
## col_corr_abdbri ~ U_landscape * U_local + (1 | plotid/location)
## Data: data_col
##
## AIC BIC logLik deviance df.resid
## 8580.3 8643.1 -4278.2 8556.3 1367
##
## Random effects:
##
## Conditional model:
## Groups Name Variance Std.Dev.
## location:plotid (Intercept) 1.219 1.104
## plotid (Intercept) 2.089 1.445
## Residual 27.437 5.238
## Number of obs: 1379, groups: location:plotid, 80; plotid, 27
##
## Dispersion estimate for gaussian family (sigma^2): 27.4
##
## Conditional model:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 23.3474 0.7447 31.354 <2e-16 ***
## U_landscapeMEDIUM -0.0595 1.0613 -0.056 0.9553
## U_landscapeHIGH -1.8578 1.0393 -1.788 0.0738 .
## U_localMEDIUM -0.6655 0.9286 -0.717 0.4736
## U_localHIGH -0.9605 0.7883 -1.218 0.2231
## U_landscapeMEDIUM:U_localMEDIUM -0.5364 1.2295 -0.436 0.6626
## U_landscapeHIGH:U_localMEDIUM 2.6753 1.2206 2.192 0.0284 *
## U_landscapeMEDIUM:U_localHIGH 0.3806 1.1208 0.340 0.7342
## U_landscapeHIGH:U_localHIGH 2.4678 1.1041 2.235 0.0254 *
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## Analysis of Deviance Table (Type III Wald chisquare tests)
##
## Response: col_corr_abdbri
## Chisq Df Pr(>Chisq)
## (Intercept) 983.0451 1 < 2e-16 ***
## U_landscape 4.1454 2 0.12585
## U_local 1.5228 2 0.46701
## U_landscape:U_local 10.8533 4 0.02826 *
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
order this table (nothing is significant when multiple comparisons are
used)
visualisation
## # A tibble: 9 × 7
## U_landscape U_local emmean SE df lower.CL upper.CL
## <fct> <fct> <dbl> <dbl> <dbl> <dbl> <dbl>
## 1 LOW LOW 23.3 0.745 1367 21.9 24.8
## 2 MEDIUM LOW 23.3 0.756 1367 21.8 24.8
## 3 HIGH LOW 21.5 0.725 1367 20.1 22.9
## 4 LOW MEDIUM 22.7 0.875 1367 21.0 24.4
## 5 MEDIUM MEDIUM 22.1 0.735 1367 20.6 23.5
## 6 HIGH MEDIUM 23.5 0.753 1367 22.0 25.0
## 7 LOW HIGH 22.4 0.729 1367 21.0 23.8
## 8 MEDIUM HIGH 22.7 0.726 1367 21.3 24.1
## 9 HIGH HIGH 23.0 0.733 1367 21.6 24.4
interaction is significant however outlier test is significant and KS
test => maybe best check trends with data_col_filt
repeat statistics but than for the filtered dataset ###statistics - data_col_filt
## Family: gaussian ( identity )
## Formula:
## col_corr_abdbri ~ U_landscape * U_local + (1 | plotid/location)
## Data: data_col_filt
##
## AIC BIC logLik deviance df.resid
## 7962.6 8025.0 -3969.3 7938.6 1325
##
## Random effects:
##
## Conditional model:
## Groups Name Variance Std.Dev.
## location:plotid (Intercept) 0.6069 0.7791
## plotid (Intercept) 1.7694 1.3302
## Residual 21.1137 4.5950
## Number of obs: 1337, groups: location:plotid, 80; plotid, 27
##
## Dispersion estimate for gaussian family (sigma^2): 21.1
##
## Conditional model:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 23.0347 0.6417 35.90 < 2e-16 ***
## U_landscapeMEDIUM -0.2022 0.9149 -0.22 0.82509
## U_landscapeHIGH -2.0578 0.8955 -2.30 0.02157 *
## U_localMEDIUM -0.9267 0.7631 -1.21 0.22456
## U_localHIGH -0.9141 0.6409 -1.43 0.15380
## U_landscapeMEDIUM:U_localMEDIUM 0.2674 1.0064 0.27 0.79049
## U_landscapeHIGH:U_localMEDIUM 3.0658 1.0006 3.06 0.00218 **
## U_landscapeMEDIUM:U_localHIGH 0.5583 0.9120 0.61 0.54044
## U_landscapeHIGH:U_localHIGH 2.1422 0.8999 2.38 0.01729 *
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## Analysis of Deviance Table (Type III Wald chisquare tests)
##
## Response: col_corr_abdbri
## Chisq Df Pr(>Chisq)
## (Intercept) 1288.6367 1 < 2.2e-16 ***
## U_landscape 6.4467 2 0.039822 *
## U_local 2.4532 2 0.293293
## U_landscape:U_local 13.9562 4 0.007436 **
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## contrast estimate SE df t.ratio p.value
## LOW - MEDIUM -0.073 0.740 1325 -0.099 0.9946
## LOW - HIGH 0.322 0.741 1325 0.434 0.9012
## MEDIUM - HIGH 0.395 0.730 1325 0.541 0.8510
##
## Results are averaged over the levels of: U_local
## P value adjustment: tukey method for comparing a family of 3 estimates
## contrast estimate SE df t.ratio p.value
## LOW LOW - MEDIUM LOW 0.2022 0.915 1325 0.221 1.0000
## LOW LOW - HIGH LOW 2.0578 0.896 1325 2.298 0.3440
## LOW LOW - LOW MEDIUM 0.9267 0.763 1325 1.214 0.9534
## LOW LOW - MEDIUM MEDIUM 0.8616 0.900 1325 0.957 0.9894
## LOW LOW - HIGH MEDIUM -0.0813 0.914 1325 -0.089 1.0000
## LOW LOW - LOW HIGH 0.9141 0.641 1325 1.426 0.8878
## LOW LOW - MEDIUM HIGH 0.5580 0.895 1325 0.624 0.9995
## LOW LOW - HIGH HIGH 0.8297 0.903 1325 0.919 0.9919
## MEDIUM LOW - HIGH LOW 1.8556 0.903 1325 2.055 0.5051
## MEDIUM LOW - LOW MEDIUM 0.7245 0.992 1325 0.730 0.9984
## MEDIUM LOW - MEDIUM MEDIUM 0.6594 0.656 1325 1.005 0.9855
## MEDIUM LOW - HIGH MEDIUM -0.2835 0.921 1325 -0.308 1.0000
## MEDIUM LOW - LOW HIGH 0.7119 0.904 1325 0.787 0.9972
## MEDIUM LOW - MEDIUM HIGH 0.3558 0.649 1325 0.548 0.9998
## MEDIUM LOW - HIGH HIGH 0.6275 0.910 1325 0.689 0.9989
## HIGH LOW - LOW MEDIUM -1.1310 0.974 1325 -1.161 0.9643
## HIGH LOW - MEDIUM MEDIUM -1.1962 0.888 1325 -1.347 0.9169
## HIGH LOW - HIGH MEDIUM -2.1391 0.647 1325 -3.305 0.0271
## HIGH LOW - LOW HIGH -1.1437 0.885 1325 -1.293 0.9335
## HIGH LOW - MEDIUM HIGH -1.4997 0.883 1325 -1.699 0.7472
## HIGH LOW - HIGH HIGH -1.2281 0.632 1325 -1.943 0.5835
## LOW MEDIUM - MEDIUM MEDIUM -0.0652 0.979 1325 -0.067 1.0000
## LOW MEDIUM - HIGH MEDIUM -1.0080 0.991 1325 -1.017 0.9843
## LOW MEDIUM - LOW HIGH -0.0126 0.747 1325 -0.017 1.0000
## LOW MEDIUM - MEDIUM HIGH -0.3687 0.974 1325 -0.379 1.0000
## LOW MEDIUM - HIGH HIGH -0.0970 0.981 1325 -0.099 1.0000
## MEDIUM MEDIUM - HIGH MEDIUM -0.9429 0.906 1325 -1.040 0.9819
## MEDIUM MEDIUM - LOW HIGH 0.0526 0.889 1325 0.059 1.0000
## MEDIUM MEDIUM - MEDIUM HIGH -0.3035 0.628 1325 -0.483 0.9999
## MEDIUM MEDIUM - HIGH HIGH -0.0319 0.895 1325 -0.036 1.0000
## HIGH MEDIUM - LOW HIGH 0.9954 0.903 1325 1.102 0.9739
## HIGH MEDIUM - MEDIUM HIGH 0.6393 0.901 1325 0.710 0.9987
## HIGH MEDIUM - HIGH HIGH 0.9110 0.657 1325 1.387 0.9028
## LOW HIGH - MEDIUM HIGH -0.3561 0.884 1325 -0.403 1.0000
## LOW HIGH - HIGH HIGH -0.0844 0.892 1325 -0.095 1.0000
## MEDIUM HIGH - HIGH HIGH 0.2717 0.890 1325 0.305 1.0000
##
## P value adjustment: tukey method for comparing a family of 9 estimates
| contrast | estimate | SE | df | t.ratio | p.value |
|---|---|---|---|---|---|
| HIGH LOW - HIGH MEDIUM | -2.13907 | 0.6471824 | 1325 | -3.305204 | 0.0271121 |
difference is between local low urbanisation level and local medium urbanisation level within highly urbanised landscapes
but still outlier test significant and ks test as well
## # A tibble: 9 × 7
## U_landscape U_local emmean SE df lower.CL upper.CL
## <fct> <fct> <dbl> <dbl> <dbl> <dbl> <dbl>
## 1 LOW LOW 23.0 0.642 1325 21.8 24.3
## 2 MEDIUM LOW 22.8 0.652 1325 21.6 24.1
## 3 HIGH LOW 21.0 0.625 1325 19.8 22.2
## 4 LOW MEDIUM 22.1 0.748 1325 20.6 23.6
## 5 MEDIUM MEDIUM 22.2 0.631 1325 20.9 23.4
## 6 HIGH MEDIUM 23.1 0.650 1325 21.8 24.4
## 7 LOW HIGH 22.1 0.626 1325 20.9 23.3
## 8 MEDIUM HIGH 22.5 0.624 1325 21.3 23.7
## 9 HIGH HIGH 22.2 0.635 1325 21.0 23.5
comparison of the outputs of modA and modA_filt, so the normal color
dataset and the filtered colour dataset
in summary on the graph a trend is visible: -in low urbanised landscapes: spiders tend to become darker with increasing urbanisation level at local scale -in medium urbanised landscapes: spiders tend to become darker with increasing urbanisation level at local scale, however spiders in the medium urbanised locations are the darkest, -in high urbanised landscapes: spiders tend to become paler with increasing urbanisation level, however the reflectance of spiders in locally high urbanised locations is roughly the same across the urbanisation levels at landscape scale. meaning that all “purple” dots are roughly the same across the landscape urbanisation gradient. It is mainly the spiders in the least urbanised location that are very dark
or rephrased: the significant interaction indicates that depending on the U_landscape context so whether it is low&medium vs high that the effect of urbanisation on the local scale is different. i.e. in low&medium landscapes the spiders in urbanised local sites are darker than there natural counterparts but in high urbanised landscapes the spiders in local urban areas are paler than there natural counterpart. However if you compare the this high-high to the others the spiders are still quite dark in a very urbanised setting compared to other ubanisation settings
only significant difference was found within the highly urbanised landscapes: namely significant difference between spiders in locally low urbanised and locally medium urbanised locations: local low urbanised spiders (very dark spiders) and local medium urbanised spiders (very light spiders) is significantly different
maybe less usefull big range of values are based on 3 point
measurements => average of this => this average then corrected so
you get % reflectance ####statistics-leaflight
## Family: gaussian ( identity )
## Formula:
## col_corr_leaflight ~ U_landscape * U_local + (1 | plotid/location)
## Data: data_col
##
## AIC BIC logLik deviance df.resid
## 9576.9 9639.6 -4776.4 9552.9 1367
##
## Random effects:
##
## Conditional model:
## Groups Name Variance Std.Dev.
## location:plotid (Intercept) 1.516 1.231
## plotid (Intercept) 3.569 1.889
## Residual 57.140 7.559
## Number of obs: 1379, groups: location:plotid, 80; plotid, 27
##
## Dispersion estimate for gaussian family (sigma^2): 57.1
##
## Conditional model:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 24.6186 0.9757 25.232 < 2e-16 ***
## U_landscapeMEDIUM -1.2312 1.3923 -0.884 0.376523
## U_landscapeHIGH -3.4297 1.3584 -2.525 0.011577 *
## U_localMEDIUM -2.0918 1.2244 -1.708 0.087549 .
## U_localHIGH -1.5851 1.0316 -1.537 0.124394
## U_landscapeMEDIUM:U_localMEDIUM 1.1766 1.6166 0.728 0.466728
## U_landscapeHIGH:U_localMEDIUM 5.5009 1.6048 3.428 0.000608 ***
## U_landscapeMEDIUM:U_localHIGH 1.1492 1.4676 0.783 0.433618
## U_landscapeHIGH:U_localHIGH 4.3086 1.4422 2.988 0.002812 **
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## Analysis of Deviance Table (Type III Wald chisquare tests)
##
## Response: col_corr_leaflight
## Chisq Df Pr(>Chisq)
## (Intercept) 636.672 1 < 2.2e-16 ***
## U_landscape 6.574 2 0.037366 *
## U_local 3.668 2 0.159775
## U_landscape:U_local 16.512 4 0.002404 **
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## # A tibble: 36 × 6
## contrast estimate SE df t.ratio p.value
## <chr> <dbl> <dbl> <dbl> <dbl> <dbl>
## 1 LOW LOW - MEDIUM LOW 1.23 1.39 1367 0.884 0.994
## 2 LOW LOW - HIGH LOW 3.43 1.36 1367 2.52 0.221
## 3 LOW LOW - LOW MEDIUM 2.09 1.22 1367 1.71 0.741
## 4 LOW LOW - MEDIUM MEDIUM 2.15 1.37 1367 1.57 0.822
## 5 LOW LOW - HIGH MEDIUM 0.0206 1.39 1367 0.0148 1
## 6 LOW LOW - LOW HIGH 1.59 1.03 1367 1.54 0.838
## 7 LOW LOW - MEDIUM HIGH 1.67 1.36 1367 1.23 0.951
## 8 LOW LOW - HIGH HIGH 0.706 1.37 1367 0.517 1.00
## 9 MEDIUM LOW - HIGH LOW 2.20 1.37 1367 1.60 0.803
## 10 MEDIUM LOW - LOW MEDIUM 0.861 1.52 1367 0.566 1.00
## # ℹ 26 more rows
| contrast | estimate | SE | df | t.ratio | p.value |
|---|---|---|---|---|---|
| HIGH LOW - HIGH MEDIUM | -3.409108 | 1.036996 | 1367 | -3.287483 | 0.0286773 |
## # A tibble: 9 × 7
## U_landscape U_local emmean SE df lower.CL upper.CL
## <fct> <fct> <dbl> <dbl> <dbl> <dbl> <dbl>
## 1 LOW LOW 24.6 0.976 1367 22.7 26.5
## 2 MEDIUM LOW 23.4 0.994 1367 21.4 25.3
## 3 HIGH LOW 21.2 0.945 1367 19.3 23.0
## 4 LOW MEDIUM 22.5 1.15 1367 20.3 24.8
## 5 MEDIUM MEDIUM 22.5 0.960 1367 20.6 24.4
## 6 HIGH MEDIUM 24.6 0.988 1367 22.7 26.5
## 7 LOW HIGH 23.0 0.952 1367 21.2 24.9
## 8 MEDIUM HIGH 23.0 0.947 1367 21.1 24.8
## 9 HIGH HIGH 23.9 0.957 1367 22.0 25.8
## Family: gaussian ( identity )
## Formula:
## col_corr_leaflight ~ U_landscape * U_local + (1 | plotid/location)
## Data: data_col_filt
##
## AIC BIC logLik deviance df.resid
## 8982.8 9045.2 -4479.4 8958.8 1325
##
## Random effects:
##
## Conditional model:
## Groups Name Variance Std.Dev.
## location:plotid (Intercept) 0.4721 0.6871
## plotid (Intercept) 3.1583 1.7772
## Residual 45.8844 6.7738
## Number of obs: 1337, groups: location:plotid, 80; plotid, 27
##
## Dispersion estimate for gaussian family (sigma^2): 45.9
##
## Conditional model:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 24.2335 0.8496 28.524 < 2e-16 ***
## U_landscapeMEDIUM -1.7177 1.2124 -1.417 0.15657
## U_landscapeHIGH -3.6748 1.1826 -3.107 0.00189 **
## U_localMEDIUM -2.3756 1.0085 -2.356 0.01849 *
## U_localHIGH -1.5222 0.8376 -1.817 0.06916 .
## U_landscapeMEDIUM:U_localMEDIUM 2.3155 1.3236 1.749 0.08022 .
## U_landscapeHIGH:U_localMEDIUM 6.1034 1.3181 4.631 3.65e-06 ***
## U_landscapeMEDIUM:U_localHIGH 1.6794 1.1923 1.409 0.15896
## U_landscapeHIGH:U_localHIGH 3.8585 1.1741 3.286 0.00101 **
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## Analysis of Deviance Table (Type III Wald chisquare tests)
##
## Response: col_corr_leaflight
## Chisq Df Pr(>Chisq)
## (Intercept) 813.6436 1 < 2.2e-16 ***
## U_landscape 9.6839 2 0.007892 **
## U_local 6.2788 2 0.043308 *
## U_landscape:U_local 24.5583 4 6.172e-05 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## # A tibble: 36 × 6
## contrast estimate SE df t.ratio p.value
## <chr> <dbl> <dbl> <dbl> <dbl> <dbl>
## 1 LOW LOW - MEDIUM LOW 1.72 1.21 1325 1.42 0.892
## 2 LOW LOW - HIGH LOW 3.67 1.18 1325 3.11 0.0499
## 3 LOW LOW - LOW MEDIUM 2.38 1.01 1325 2.36 0.310
## 4 LOW LOW - MEDIUM MEDIUM 1.78 1.19 1325 1.49 0.858
## 5 LOW LOW - HIGH MEDIUM -0.0530 1.21 1325 -0.0438 1.00
## 6 LOW LOW - LOW HIGH 1.52 0.838 1325 1.82 0.671
## 7 LOW LOW - MEDIUM HIGH 1.56 1.18 1325 1.32 0.925
## 8 LOW LOW - HIGH HIGH 1.34 1.19 1325 1.12 0.971
## 9 MEDIUM LOW - HIGH LOW 1.96 1.19 1325 1.64 0.783
## 10 MEDIUM LOW - LOW MEDIUM 0.658 1.32 1325 0.499 1.00
## # ℹ 26 more rows
| contrast | estimate | SE | df | t.ratio | p.value |
|---|---|---|---|---|---|
| LOW LOW - HIGH LOW | 3.674759 | 1.1826288 | 1325 | 3.107281 | 0.0499430 |
| HIGH LOW - HIGH MEDIUM | -3.727751 | 0.8470395 | 1325 | -4.400918 | 0.0003965 |
## # A tibble: 9 × 7
## U_landscape U_local emmean SE df lower.CL upper.CL
## <fct> <fct> <dbl> <dbl> <dbl> <dbl> <dbl>
## 1 LOW LOW 24.2 0.850 1325 22.6 25.9
## 2 MEDIUM LOW 22.5 0.866 1325 20.8 24.2
## 3 HIGH LOW 20.6 0.823 1325 18.9 22.2
## 4 LOW MEDIUM 21.9 0.994 1325 19.9 23.8
## 5 MEDIUM MEDIUM 22.5 0.833 1325 20.8 24.1
## 6 HIGH MEDIUM 24.3 0.864 1325 22.6 26.0
## 7 LOW HIGH 22.7 0.825 1325 21.1 24.3
## 8 MEDIUM HIGH 22.7 0.821 1325 21.1 24.3
## 9 HIGH HIGH 22.9 0.839 1325 21.2 24.5
result: trends visualy the same (some change for medium
landscapes)
####statistics leafdark
## Family: gaussian ( identity )
## Formula:
## col_corr_leafdark ~ U_landscape * U_local + (1 | plotid/location)
## Data: data_col
##
## AIC BIC logLik deviance df.resid
## 8109.8 8172.6 -4042.9 8085.8 1367
##
## Random effects:
##
## Conditional model:
## Groups Name Variance Std.Dev.
## location:plotid (Intercept) 0.3624 0.602
## plotid (Intercept) 1.4393 1.200
## Residual 19.7646 4.446
## Number of obs: 1379, groups: location:plotid, 80; plotid, 27
##
## Dispersion estimate for gaussian family (sigma^2): 19.8
##
## Conditional model:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 15.6023 0.5776 27.013 <2e-16 ***
## U_landscapeMEDIUM -0.3606 0.8239 -0.438 0.6616
## U_landscapeHIGH -1.3140 0.8046 -1.633 0.1025
## U_localMEDIUM -0.3613 0.6881 -0.525 0.5996
## U_localHIGH -0.6943 0.5757 -1.206 0.2278
## U_landscapeMEDIUM:U_localMEDIUM 0.2561 0.9063 0.283 0.7775
## U_landscapeHIGH:U_localMEDIUM 2.0352 0.8990 2.264 0.0236 *
## U_landscapeMEDIUM:U_localHIGH 0.8048 0.8195 0.982 0.3260
## U_landscapeHIGH:U_localHIGH 1.0246 0.8043 1.274 0.2027
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## Analysis of Deviance Table (Type III Wald chisquare tests)
##
## Response: col_corr_leafdark
## Chisq Df Pr(>Chisq)
## (Intercept) 729.7089 1 < 2e-16 ***
## U_landscape 2.8640 2 0.23883
## U_local 1.4549 2 0.48314
## U_landscape:U_local 8.1710 4 0.08551 .
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## # A tibble: 36 × 6
## contrast estimate SE df t.ratio p.value
## <chr> <dbl> <dbl> <dbl> <dbl> <dbl>
## 1 LOW LOW - MEDIUM LOW 0.361 0.824 1367 0.438 1.00
## 2 LOW LOW - HIGH LOW 1.31 0.805 1367 1.63 0.787
## 3 LOW LOW - LOW MEDIUM 0.361 0.688 1367 0.525 1.00
## 4 LOW LOW - MEDIUM MEDIUM 0.466 0.811 1367 0.575 1.00
## 5 LOW LOW - HIGH MEDIUM -0.360 0.822 1367 -0.438 1.00
## 6 LOW LOW - LOW HIGH 0.694 0.576 1367 1.21 0.955
## 7 LOW LOW - MEDIUM HIGH 0.250 0.805 1367 0.311 1.00
## 8 LOW LOW - HIGH HIGH 0.984 0.809 1367 1.22 0.953
## 9 MEDIUM LOW - HIGH LOW 0.953 0.812 1367 1.17 0.962
## 10 MEDIUM LOW - LOW MEDIUM 0.000623 0.895 1367 0.000696 1
## # ℹ 26 more rows
| contrast | estimate | SE | df | t.ratio | p.value |
|---|---|---|---|---|---|
| HIGH LOW - HIGH MEDIUM | -1.673927 | 0.5793073 | 1367 | -2.889532 | 0.0920716 |
## # A tibble: 9 × 7
## U_landscape U_local emmean SE df lower.CL upper.CL
## <fct> <fct> <dbl> <dbl> <dbl> <dbl> <dbl>
## 1 LOW LOW 15.6 0.578 1367 14.5 16.7
## 2 MEDIUM LOW 15.2 0.588 1367 14.1 16.4
## 3 HIGH LOW 14.3 0.560 1367 13.2 15.4
## 4 LOW MEDIUM 15.2 0.675 1367 13.9 16.6
## 5 MEDIUM MEDIUM 15.1 0.569 1367 14.0 16.3
## 6 HIGH MEDIUM 16.0 0.585 1367 14.8 17.1
## 7 LOW HIGH 14.9 0.564 1367 13.8 16.0
## 8 MEDIUM HIGH 15.4 0.561 1367 14.3 16.5
## 9 HIGH HIGH 14.6 0.567 1367 13.5 15.7
assumptions of the statistics not ok
####statisticks - filtered leafdark
## Family: gaussian ( identity )
## Formula:
## col_corr_leafdark ~ U_landscape * U_local + (1 | plotid/location)
## Data: data_col_filt
##
## AIC BIC logLik deviance df.resid
## 7666.8 7729.2 -3821.4 7642.8 1325
##
## Random effects:
##
## Conditional model:
## Groups Name Variance Std.Dev.
## location:plotid (Intercept) 0.2497 0.4997
## plotid (Intercept) 1.4570 1.2071
## Residual 17.0460 4.1287
## Number of obs: 1337, groups: location:plotid, 80; plotid, 27
##
## Dispersion estimate for gaussian family (sigma^2): 17
##
## Conditional model:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) 15.4502 0.5552 27.829 < 2e-16 ***
## U_landscapeMEDIUM -0.3826 0.7915 -0.483 0.62883
## U_landscapeHIGH -1.4505 0.7743 -1.873 0.06103 .
## U_localMEDIUM -0.6700 0.6329 -1.059 0.28974
## U_localHIGH -0.7624 0.5267 -1.448 0.14772
## U_landscapeMEDIUM:U_localMEDIUM 0.8021 0.8315 0.965 0.33469
## U_landscapeHIGH:U_localMEDIUM 2.4467 0.8268 2.959 0.00308 **
## U_landscapeMEDIUM:U_localHIGH 0.9169 0.7498 1.223 0.22139
## U_landscapeHIGH:U_localHIGH 0.9566 0.7388 1.295 0.19537
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## Analysis of Deviance Table (Type III Wald chisquare tests)
##
## Response: col_corr_leafdark
## Chisq Df Pr(>Chisq)
## (Intercept) 774.4422 1 < 2e-16 ***
## U_landscape 3.7905 2 0.15028
## U_local 2.3088 2 0.31524
## U_landscape:U_local 11.2784 4 0.02361 *
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## # A tibble: 36 × 6
## contrast estimate SE df t.ratio p.value
## <chr> <dbl> <dbl> <dbl> <dbl> <dbl>
## 1 LOW LOW - MEDIUM LOW 0.383 0.792 1325 0.483 1.00
## 2 LOW LOW - HIGH LOW 1.45 0.774 1325 1.87 0.632
## 3 LOW LOW - LOW MEDIUM 0.670 0.633 1325 1.06 0.980
## 4 LOW LOW - MEDIUM MEDIUM 0.250 0.778 1325 0.322 1.00
## 5 LOW LOW - HIGH MEDIUM -0.326 0.791 1325 -0.412 1.00
## 6 LOW LOW - LOW HIGH 0.762 0.527 1325 1.45 0.879
## 7 LOW LOW - MEDIUM HIGH 0.228 0.774 1325 0.295 1.00
## 8 LOW LOW - HIGH HIGH 1.26 0.781 1325 1.61 0.800
## 9 MEDIUM LOW - HIGH LOW 1.07 0.781 1325 1.37 0.910
## 10 MEDIUM LOW - LOW MEDIUM 0.287 0.855 1325 0.336 1.00
## # ℹ 26 more rows
| contrast | estimate | SE | df | t.ratio | p.value |
|---|---|---|---|---|---|
| HIGH LOW - HIGH MEDIUM | -1.776658 | 0.5324686 | 1325 | -3.336643 | 0.0244914 |
| HIGH MEDIUM - HIGH HIGH | 1.582450 | 0.5416742 | 1325 | 2.921405 | 0.0845517 |
## # A tibble: 9 × 7
## U_landscape U_local emmean SE df lower.CL upper.CL
## <fct> <fct> <dbl> <dbl> <dbl> <dbl> <dbl>
## 1 LOW LOW 15.5 0.555 1325 14.4 16.5
## 2 MEDIUM LOW 15.1 0.564 1325 14.0 16.2
## 3 HIGH LOW 14.0 0.540 1325 12.9 15.1
## 4 LOW MEDIUM 14.8 0.642 1325 13.5 16.0
## 5 MEDIUM MEDIUM 15.2 0.546 1325 14.1 16.3
## 6 HIGH MEDIUM 15.8 0.563 1325 14.7 16.9
## 7 LOW HIGH 14.7 0.541 1325 13.6 15.7
## 8 MEDIUM HIGH 15.2 0.539 1325 14.2 16.3
## 9 HIGH HIGH 14.2 0.549 1325 13.1 15.3
combination of the results
result: trends visualy the same (some change for medium
landscapes) using the filterd dataset : data_col_filt (seen on the right
side of the graph): HIGH LOW - HIGH MEDIUM -1.7767 0.532 1325 -3.337
0.0245
with data_col : HIGH LOW - HIGH MEDIUM -1.673927 0.579 1367 -2.890 0.0921 (marginaly signif) in the model interaction was marginally significant as well
When exploring leaflight and leafdark in roughly the same trends as for abdomen brightness
#not necessary for report
##questions about statistical tests
rptR and lme4
question: not completely clear how to use rptR for our dataset
found in literature:
https://onlinelibrary.wiley.com/doi/10.1111/j.1469-185X.2010.00141.x Adjusted repeatability controlling for fixed effects (U_landscape and U_local), was then calculated using the rptR package in R (Nakagawa & Schielzeth, 2013; Stoffel et al., 2017). estimating repeatability (intra-class correlation) and confidence intervals (C.I) from Gaussian data (Stoffel et al. 2017) Unadjusted repeatability measures the between-individual variation in a given …response variable……, while adjusted repeatability controls for fixed effects that could influence ….response variable….., either because they explain between or within individual components of variation. For both adjusted and unadjusted repeatability, we included ….location…. as a random effect info : https://cran.r-project.org/web/packages/rptR/vignettes/rptR.html
I have set this code chunk as eval=FALSE so they will not be evaluated on knit
remark lme4 gives different output, because is based on REML REML necessary for small sample sizes for model comparisons ML is needed. for now I used the ML method
I have set these chunks as eval=FALSE so they will not be evaluated on knit
however the degrees of freedom when asking emmeans is different kenward-roger approximation only applied to REML models Satterthwaite approximation can be applied to ML and REML models
I have set these chunks as eval=FALSE so they will not be evaluated on knit
performance package is developed for models via lme4 ! I have set these chunks as eval=FALSE so they will not be evaluated on knit